• Journal of Applied Optics
  • Vol. 41, Issue 2, 288 (2020)
Qingjiang CHEN1, Xiaohan SHI1,*, and Yuzhou CHAI2
Author Affiliations
  • 1College of Science, Xi’an University of Architecture and Technology, Xi’an 710055, China
  • 2Xi’an Institute of Space Radio Technology, Xi’an 710000, China
  • show less
    DOI: 10.5768/JAO202041.0202001 Cite this Article
    Qingjiang CHEN, Xiaohan SHI, Yuzhou CHAI. Image denoising algorithm based on wavelet transform and convolutional neural network[J]. Journal of Applied Optics, 2020, 41(2): 288 Copy Citation Text show less
    Structure diagram of proposed network
    Fig. 1. Structure diagram of proposed network
    Flow chart of denoising algorithm
    Fig. 2. Flow chart of denoising algorithm
    Comparison of denoising performance with different algorithms
    Fig. 3. Comparison of denoising performance with different algorithms
    Comparison of denoising performance with different algorithms
    Fig. 4. Comparison of denoising performance with different algorithms
    Image denoising results of real noisy images based on proposed algorithm
    Fig. 5. Image denoising results of real noisy images based on proposed algorithm
    Image denoising results of remote sensing image based on proposed algorithm
    Fig. 6. Image denoising results of remote sensing image based on proposed algorithm
    MethodsstarfishairplanemonarchparrotbridgeflowersAverage
    MF25.3124.4625.3724.8423.9525.6424.93
    ST24.9124.0724.6924.5023.8424.9124.49
    EPLL28.5128.6029.3928.9526.4728.9828.48
    NCSR28.7628.4529.4328.8626.2928.7728.43
    WNNM21.8922.1522.1721.6122.0822.4922.07
    Ours29.6928.5629.5329.5026.6229.6928.93
    Table 1. PSNR of experimental results with different algorithms dB
    MethodsstarfishairplanemonarchparrotbridgeflowersAverage
    MF0.698 00.591 20.702 70.609 00.622 30.640 30.643 9
    ST0.670 00.551 10.649 90.569 70.640 90.603 50.614 1
    EPLL0.851 40.863 00.902 00.852 40.761 80.833 40.844 0
    NCSR0.852 90.857 60.903 00.848 70.742 30.822 70.837 9
    WNNM0.625 50.689 30.750 60.709 10.465 00.621 80.643 6
    Ours0.862 80.864 90.899 90.853 80.729 60.834 40.840 9
    Table 2. SSIM of experimental results with different algorithms
    MethodsMFSTEPLLNCSRWNNM本文
    Average time/s2.970.2253.53313.73240.561.46
    Table 3. Average time of experimental results with different algorithms
    MethodsC.manlenapepperswomanbabybirdAverage
    MF20.5321.5421.6128.4821.9621.8722.67
    ST19.7420.0920.1820.2920.5220.4220.21
    EPLL26.0928.6129.0527.0228.6528.0627.91
    NCSR26.1528.9129.2227.1328.9328.0028.06
    WNNM21.2225.9625.5322.3026.4023.5924.17
    Ours26.6329.1929.5927.4229.0228.3628.37
    Table 4. PSNR of experimental results with different algorithms dB
    MethodsC.manlenapepperswomanbabybirdAverage
    MF0.334 60.325 70.316 50.416 90.342 30.409 90.357 7
    ST0.309 70.273 00.261 40.371 10.284 20.342 00.306 9
    EPLL0.775 40.787 00.778 20.825 30.779 90.796 70.790 4
    NCSR0.783 40.803 50.790 70.842 90.792 50.814 30.804 6
    WNNM0.655 80.747 10.736 40.733 60.722 10.700 30.715 9
    Ours0.783 40.797 90.789 10.851 70.792 40.833 70.808 0
    Table 5. SSIM of experimental results with different algorithms
    Qingjiang CHEN, Xiaohan SHI, Yuzhou CHAI. Image denoising algorithm based on wavelet transform and convolutional neural network[J]. Journal of Applied Optics, 2020, 41(2): 288
    Download Citation